Core Web Vitals measure how a real user experiences your page loading, responding, and settling. Since March 2024, INP replaced FID as the responsiveness metric, and all three vitals are confirmed inputs to Google's page experience signals. Getting them right is table stakes for competitive queries.
This guide covers the current thresholds, how to read your data, and the specific fix for each metric.
The 2026 thresholds
| Metric | Good | Needs improvement | Poor |
|---|---|---|---|
| LCP (Largest Contentful Paint) | under 2.5s | 2.5s to 4.0s | over 4.0s |
| INP (Interaction to Next Paint) | 200ms or less | 200ms to 500ms | over 500ms |
| CLS (Cumulative Layout Shift) | 0.1 or less | 0.1 to 0.25 | over 0.25 |
Two 2026 notes. First, INP is now one of the three official Core Web Vitals and a live ranking input since it replaced FID on 12 March 2024. Second, guidance has tightened around LCP: while 2.5s remains the documented "good" ceiling, faster sites are treated more favorably, and practitioners are targeting sub-2.0s LCP to stay comfortably clear of the "needs improvement" band. Aim for 2.0s, not 2.4s.
You pass a metric for a URL group when at least 75% of real-user visits fall in the "good" range. All three must pass for the group to be "good" overall.
Field data vs lab data
Two data sources, and they answer different questions.
Field data comes from real users (the Chrome UX Report). This is what Google uses for the page experience signal. If it is not in the field, it does not count for ranking.
Lab data comes from a synthetic test (like Lighthouse) in a controlled environment. Useful for reproducing and debugging a problem, but it will not match field numbers because real users have slower devices and networks.
Audit rule: diagnose direction from field data, debug the cause in lab data.
Fixing LCP
LCP measures when the largest visible element (usually a hero image, heading, or video poster) finishes rendering. Slow LCP is almost always render-blocking resources or a slow-loading hero asset.
Common fixes:
- Preload the LCP image with fetchpriority="high" and stop lazy-loading it.
- Serve modern image formats (WebP, AVIF) and correctly sized responsive images.
- Cut render-blocking CSS and JavaScript; inline critical CSS, defer the rest.
- Improve server response time (TTFB); a slow origin caps how fast LCP can ever be.
- Use a CDN so the largest asset loads from a nearby edge.
Fixing INP
INP measures the delay between a user interaction (tap, click, keypress) and the next visual update, across the whole session, not just the first interaction. High INP means the main thread is busy when the user tries to interact.
Common fixes:
- Break up long JavaScript tasks so the main thread is not blocked for 50ms-plus stretches.
- Defer or remove non-essential third-party scripts (chat widgets, heavy analytics, tag managers).
- Yield to the main thread during expensive work so the UI can paint.
- Avoid large DOM updates on every interaction; batch and debounce them.
- Replace heavy client-side rendering with server rendering where you can.
Fixing CLS
CLS measures unexpected layout shifts as the page loads. The classic cause is content moving because something above it loaded late.
Common fixes:
- Set explicit width and height (or aspect-ratio) on images, videos, and iframes.
- Reserve space for ads, embeds, and banners so they do not push content down.
- Preload fonts and use font-display: optional or swap carefully to avoid reflow.
- Never insert content above existing content unless it is in response to a user action.
An audit workflow that holds up
Pull field data per URL group and find which metric fails the 75% bar.
Reproduce the failing metric in a lab tool to see the cause.
Fix the single biggest contributor first; do not chase all three at once.
Deploy, then wait for field data to update (it lags because it is a 28-day rolling window).
Confirm the group moved into "good" before moving to the next issue.
The field-data lag trips people up. You cannot fix INP on Monday and confirm it on Tuesday. Fix, ship, and check the trend over the following weeks.
FAQ
Is INP a ranking factor in 2026?
Yes. INP replaced FID on 12 March 2024 and is one of the three Core Web Vitals that feed Google's page experience signals. The "good" threshold is 200ms or less.
Has the LCP threshold changed?
The documented "good" ceiling remains 2.5s, but the practical bar has tightened toward 2.0s. Sites in the 2.0s to 2.5s range are increasingly treated as room for improvement, so target sub-2.0s.
Why does my Lighthouse score look great but field data is poor?
Lighthouse is a lab test on a controlled machine. Field data reflects real users on slower devices and networks. Google ranks on field data, so trust it over lab scores.
What does the 75% rule mean?
A URL group passes a metric only when at least 75% of real-user visits land in the "good" range. All three metrics must pass for the group to be assessed as good.
How long until fixes show up?
Field data uses a 28-day rolling window, so improvements appear gradually over weeks, not instantly. Ship the fix and watch the trend.
Do Core Web Vitals matter for AI search?
Indirectly. Fast, stable pages are crawled more efficiently and rank better in traditional search, which feeds the same content pool AI answer engines draw from.
See your Core Web Vitals in context, not in isolation. A free SEO AEO Specialist crawl (€0, up to 50 pages) flags speed and stability issues alongside your crawl and indexing errors, so you fix the page as a whole. Pull a one-time report for €9, or monitor a domain continuously for €19/mo. Start free at SEO AEO Specialist or return to the SEO audit hub.
Why Is My Site Not Indexed? Indexation Troubleshooting Guide | SEO AEO Specialist
See where you stand
SEO AEO Specialist runs a free AI-visibility audit and hands you the exact fixes. One-off report €9, unlimited €19/mo.
Run your free audit →